Quality Engineer AI First

Quality Engineer (AI-first Testing)

About the role

You will be the first AI-first quality engineer at Verdantis, testing an AI-native platform MRO360 and our MDM product that large plants in energy, manufacturing and mining use to run their spare parts, equipment and material master data. The product makes decisions, not just screens: it recommends stock levels, flags duplicate materials and classifies parts. Testing means testing both the software and the judgement.

We do not want a manual tester with a 400-row spreadsheet, nor a classic automation engineer maintaining brittle scripts. We want someone early in their career who already uses AI to generate test cases, write and repair automation, explore edge cases and read logs and who wants to find out how far that can go. You will set the pattern for how quality is done here.

What you do

Functional testing

• Own test coverage for new features across MRO360 and MDM: read the story, generate the test cases with AI, prune and sharpen them yourself, and execute them before the sprint closes.
• Test the AI behavior, not just the UI: build evaluation sets for recommendations (stock levels, duplicate detection, classification), define what correct means with the product manager and data scientists, and track accuracy release over release.
• Reproduce customer-reported defects with real-shaped data (SAP material master, equipment, BOM extracts), write them up so a developer can fix them in one pass, and verify the fix.

Platform testing

• Build and maintain automated regression suites (API and UI) written with AI, reviewed and owned by you that run on every build.
• Run integration tests against SAP-style data flows: flat-file loads, OData/IDoc payloads, error handling, idempotency on re-runs.
• Test what enterprise customers security and IT teams will ask about: role-based access, multi-tenant data isolation, performance at 200K+ SKUs, audit logs.

Raising the bar

• Keep a living record of which AI tools and prompts work for which testing tasks and share it with engineering, so test generation becomes part of how developers work.
• Report release readiness in one page the Engineering Lead and the Delivery Head can act on: what was tested, what was not, what is risky.
• Push back on shipping when the evidence says no.

What we are looking for

• 1-3 years in software testing or development, or a strong fresher with projects that show it. Attitude and tooling fluency matter more than years.
• Uses AI tools daily and can show it: a test suite generated and refined with an LLM, an automation script debugged with Copilot or Claude, a prompt library built for test design. We will ask for a live demonstration.
• Can write code: Python or JavaScript/TypeScript well enough to build and repair automation (Playwright, Selenium, pytest, Postman/REST clients) without waiting for a developer.
• Knows the basics of good testing: boundary cases, equivalence classes, negative paths, test data design and can explain why an AI-generated test case is weak.
• Comfortable with SQL for checking data loads and results against the database.
• Curious about the domain: willing to learn what a material master, a reorder point or an equipment BOM is, because the bugs that matter are domain bugs.
• Writes a clear defect report and a clear release note. Argues with evidence, not opinion.
• Based in or willing to relocate to Mumbai.

Nice to have

• Exposure to testing LLM-based features: prompt regression, evaluation datasets, hallucination and consistency checks.
• Experience with CI pipelines (GitHub Actions or similar) and running tests in them.
• Basic understanding of cloud deployments and multi-tenant SaaS.
• Any exposure to SAP, EAM or supply chain software is useful, not required.
• Performance or load testing with k6, JMeter or Locust.

Who this role is not for

• Someone who wants a test plan handed to them and executes it line by line.
• Someone who sees AI as a shortcut to avoid understanding the product. The tool generates; you are accountable.
• Someone who wants to stay in one layer. This role spans UI, API, data and model behaviour.
• Someone who is uncomfortable telling a senior developer their feature is not ready.